A comment from the coordinator, Deep Breaths blog post by Dr. Mark Ansermino, upcoming events, and PIN member and partner updates for February 2018.
A Comment from the Coordinator
Our webpage will better serve our growing membership and give you, our network members, the power to easily access important and up-to-date information about pneumonia from anywhere and at anytime. Welcome to our new webpage!
Deep Breaths: Blogs from Network Members
Can artificial intelligence help win the fight against pneumonia?
by Dr. Mark Ansermino, University of British Columbia
We use artificial intelligence (AI) almost every day. It is present in our phones, computers, cars and homes. So what about healthcare? Over the last 2-3 years there has been an increasing interest and investment in AI for healthcare, specifically through the use of smart applications. However, there are many concerns about the impact of AI. Can we trust computers to make our healthcare decisions? Will the adoption of AI result in the loss of human interaction being at the heart of healthcare? Or result in me losing my job?
Pneumonia, a highly treatable disease, is still the biggest killer of children globally. While vaccines have been shown to reduce the risk of pneumonia, they will not completely eliminate the deaths. What we need is better ways to identify children who need lifesaving treatment. Antibiotics are inexpensive and increasingly available in low resource settings where the deaths from pneumonia are occurring.
Pneumonia diagnosis is challenging
The gold standard for pneumonia diagnosis is a positive culture from tissue removed from an infected lung. This is highly invasive and rarely performed. Instead, we rely on a combination of clinical observations and tests to differentiate between a child with a mild upper respiratory tract infection and those with life-threatening pneumonia. Recent evidence has called into question the reliability of the commonly used methods for identifying pneumonia in children. The chest x-ray has been used as an important tool for identifying pneumonia, and even used as the gold standard in many clinical studies. However, the interpretation of the chest x-ray can vary significantly among clinical experts. Many of the clinical signs used to diagnose pneumonia are unreliable. Often even respiratory rate and chest indrawing, the current recommendations for the identification of pneumonia by the World Health Organization, will misidentify a number of critically ill children.
How can AI help?
AI is especially powerful when combining multiple pieces of information such as the symptoms, signs and risk factors obtained from a child. Assessing this combination of data is difficult even for the most expert clinician, making it even more unreliable when done by less skilled frontline health workers. The solution is to combine symptoms, signs and risk factors using AI to produce a diagnosis. The processing of this data has been proven to be a trivial task for even the cheapest mobile device.
What do we need for AI to work?
There are two major barriers to the introduction of AI for pneumonia diagnosis. The biggest is the trust gap between computers and humans (patients and clinicians). This can be addressed with transparent processes of deployment and evaluation. The second challenge is the difficulty in providing an accurate, definitive outcome required to train the AI. However, recent advances in molecular diagnostics will advance the development of the use of AI for diagnosis. Rather than using these expensive and time-consuming techniques as a point of care diagnostic tool that requires use of valuable resources on every patient, molecular diagnostic techniques can be used to train and develop AI algorithms. These AI algorithms can then be run on non-specialized equipment, such as an everyday mobile phone, to diagnosis diseases such as pneumonia. This will be especially valuable in settings where resources are limited.
The use of AI to combine and integrate readily available clinical information holds the promise to improve the diagnosis of pneumonia. The use of the algorithms within a mobile application can even be done by front-line health workers with limited clinical training which is especially important in settings where expertise is limited. This will allow for the timely administration of antibiotics to children at risk of dying from pneumonia, while limiting exposure to unnecessary antibiotic administration for those who do not require it.
11th International Symposium on Pneumococci & Pneumococcal Diseases (ISPPD).
Melbourne, Australia. April 15 – 19, 2018.
ISPPD provides an unparalleled international forum to bring together scientific, clinical and public health experts, practitioners, and students to present, discuss, challenge and learn from the latest work and advancements on pneumococcal disease. Want to attend the conference? Register here.
36th Annual Meeting of the European Society for Paediatric Infectious Diseases (ESPID).
Malmö, Sweden. May 28 – June 2, 2018.
Building on the success and innovation of previous meetings, ESPID 2018 will feature a wide range of sessions and learning opportunities showcasing the latest research and developments in the field hosted by leading international experts.
Register to attend here.
Pneumonia Innovations Network Member and Partner Updates
Innovative antibiotic gets recognition from Johnson & Johnson's "GenH Challenge"
Child wasting (severe, recent episodes of malnutrition) is one of the leading risks of death from pneumonia. Sangwei Lu, a scientist from the University of California Berkeley, has developed "NutMox," a mix of the recommended child-friendly antibiotic amoxicillin with a ready-to-eat therapeutic food allowing malnourished children being treated for pneumonia to also be treated for underlying malnutrition. The subsequent partnership with Michael Chansler from PATH has accelerated its development to market.
This collaboration has been recognized as one of six winners of Johnson & Johnson's "Gen H" Challenge grant. We congratulate Sangwei Lu and Michael Chansler, both committed members of the Pneumonia Innovations Network, and wish them success in the next stages of development.
Read more here.
Join our LinkedIn page to receive additional updates and announcements!
Have more to say?